Figure 3. Cursor pointing (A), ‘Prince’ pointing (B), and view pointing (C).
Figure 3 illustrates, in 1D space, the three possible cases for a pointing task. In case A, a point
cursor must be moved to reach a target, specified by an interval (an area in 2D space). In case
B, symmetrical to A, an interval cursor (an area cursor in 2D space) must be moved so as to
eventually include a point target (Kabbash and Buxton, 1995). Cases A and B differ only in
their arbitrary frames of reference (Mottet et al., 2001). W is a measure of the target in one
case and of the area cursor in the other, but the computation of the ID is identical. Case C,
which seems not to have been explicitly considered in the literature, generalizes cases A and
B: a mobile interval (or area cursor in 2D) must be moved so as to overlap the target interval
(or target area in 2D). View pointing falls into case C, where the mobile interval is the view
that must be moved so that the target enters it and becomes visible.
The quantification of task difficulty is just as straightforward in case C as it is in the other two
cases. Target distance D is the distance separating the centers of the two intervals. Target
width W can be quantified in one of two ways, depending on the goal pursued in pointing. If
the user only wants to select an object, then an overlap will suffice, in which case W must be
computed as the sum of the width of the mobile interval W1 and the width of the stationary
interval W2:
ID=log2
(1)
If, however, the user wants to reach some target region to see it, as is often the case with view
pointing, then the larger interval must include the smaller interval, and W becomes the
absolute-value difference between W1 and W2:
ID g WW
A
+1
)
(2)
Either way, Fitts’ law can be assessed in view pointing just as easily as it has been in ordinary
cursor pointing. The concept of view pointing, associated with Equations 1 and 2, shows that
Fitts’ law can serve not only to model classic pointing with a cursor, but also document
navigation—obviously a broad class of user activities in HCI.
3. MODELING POINTING IN MULTISCALE
ELECTRONIC WORLDS
Let us now consider the task of pointing to a target in a multiscale information space. First we
introduce Furnas and Bederson’s (1995) space-scale diagrams, which help us to understand
multiscale documents and multiscale navigation. We then present several multiscale pointing
techniques.
3.1. Space-Scale Diagrams
Let us consider the 2D document in Figure 4 (left), with its origin at the intersection of the x
and y axes, and let us represent it at several levels of magnification. The representation at
scale 1 corresponds to the original document. The representation at scale s, noted Rs,
corresponds to the original document magnified by a factor s, that is, a point at position (x, y)
in the original document is at position (s.x, s.y) in the Rs.